Ethical considerations
Gen AI output can be really useful and creative, but it requires you to evaluate the output and the ethical context of the tool and data.
The datasets behind tools are often scraped from the open web for data there may not be much academic, or scholarly content as often that is behind paywalls on academic databases.
The ownership of the tools will often differ and may affect the content output. Some tools are intended to meet customer needs rather than provide information or 'truth', and their output and data may reflect the ideological stance of their owner.
Check FedCite or the pages on this guide for advice on how to cite or acknowledge your use of these tools.
Remember that submitting work as your own, when created by another person or AI (ChatGPT or other generative AI) without authorisation and acknowledgement is Academic Misconduct, which may result in disciplinary action. If you need extra support you can access this from many areas of the University to help them to succeed via: Studiosity, Turnitin (Text matching), Writing Centre, ASK Desk on campus and the libraries.
Ownership of the tool and who developed it can have a bearing on the filters, algorithms, and data the tool is learning from. Some tools are developed to meet customer expectations, not to provide accurate or credible results or scholarly content.
The training sources used by the GenAI tool may break copyright because these sources may have been used without asking the copyright owner’s permission. Also, your prompts may encourage reuse and wider dissemination of these copyrighted sources and thus the copyright owner never gets credit for their intellectual property. So, do not upload copyright material to summarise, e.g., PDFs, images, audio, or visual recordings including sources created by your teacher for Moodle.
Can you locate information about what data the tools are learning from? The datasets or limitations? The year range of that data?
Tools trained on open web data will reflect the data of the open web –including biases, misinformation, and fact.
Tools trained on open web data will reflect the data of the open web –including biases (e.g., gender or racial), misinformation and fact.
The output of the tool can change even when using the same prompts. One user may get a different response than another and a different response over time.
Don’t upload any personal information, as your content may be used for learning for the AI tool. The output of the tool can change even when using the same prompts. One user may get a different response than another and a different response over time.
Tools which may be free today may incur a licence fee in the future.
